Charles E. Molnar

Born 14 March 1935, Newark NJ; Died 13 December 1996, Sunnyvale CA

Co-developer of the Laboratory Instrument Computer (LINC) for laboratory use and forerunner of the personal computer; pioneer in asynchronous circuit and system design; helped identify non-linear and active-system properties of the cochlea.

Education: B.S. and M.S. Rutgers University, 1956 and 1957; D.Sc. MIT, 1966, all in Electrical Engineering.

Professional Experience: Washington University, St. Louis, 1965-1995, where he was Professor of Biomedical Computing, Physiology, Electrical Engineering, and Computer Science. He was founder of the Institute for Biomedical Computing at Washington University and its director from 1983 to 1992, and director of the Sensory Biophysics Laboratory and the Computer Systems Laboratory at Washington University; Senior Research Fellow, Sun Microsystems, Inc., 1995-1996; Director of the Science Office, Sun Microsystems, Inc., 1995-1996.

Honors and Awards: Javits Neuroscience Investigator Award -- 1985; Member of the Board of Regents of the National Library of Medicine -- 1980-1984; NIH Director's Certificate for development (with W.A. Clark) of the Laboratory INstrument Computer (LINC); A best paper award for bridging theory and practice was named after Charles E. Molnar at the Async97 workshop (http://www.win.tue.nl/cs/pa/async97); Visiting Professor at or extended visits to University of Chile, Caltech, CMU, UNC Chapel Hill, Eindhoven University of Technology, and University of Waterloo.

In 1962 Molnar co-developed MIT's Laboratory INstrument Computer, the LINC, with Wesley A. Clark. The LINC featured a keyboard plus alphanumeric/graphical display unit for interactive use, together with a block-addressed tape unit -- operated like today's diskette units -- with pockettably-small reels for the capture of both data (in either digital or analog form) and programs that were prepared on-line by the user. About twenty of these so-called "classic" LINCs were kit-assembled at MIT in 1963, twelve of them by biomedical research scientists specially selected from various institutions throughout the United States, who then returned with working computers to their own laboratories for evaluation of the instrument as a research tool. The Digital Equipment Corporation and other groups subsequently manufactured about 60 classic LINCs, primarily in assembled form; later variants appeared as the LINC-8, the PDP-12, and the microLINC-100 & microLINC-300 (the latter manufactured in early integrated circuit form by the SPEAR Corporation). Including all variants, about 1200 LINCs were at work around the world by the end of the decade. The last of the "classics" was retired in 1996 after more than thirty years of service. Relatively small and inexpensive for its day and specifically designed for individual use, the classic LINC is acknowledged by the IEEE Computer Society to be the first personal computer.

The LINC was used in Molnar's dissertation work on the relationship between audio signals and neural responses, and widely used in biomedical research laboratories. It was also central to much of his later pioneering work in auditory physiology. In 1966, Molnar was collaborating with Russell R. Pfeiffer in the establishment of the Sensory Biophysics Laboratory (SBL) in the department of physiology of the WU medical school. This laboratory investigated the functioning of the cochlea, the sensory organ of hearing. On the untimely death of Pfeiffer in 1975, Molnar took over direction of SBL. The primary contribution of SBL was promoting the now accepted notion that the cochlea is a non-linear and active system (as recently as the 1980s, the cochlea was considered by most to be a linear passive system). The laboratory was also unique in its emphasis on using the response of the auditory nerve as a non-invasive measure of the functioning of the cochlea.

With support from NIH and DARPA, and under the direction of Clark and Molnar, the design and fabrication of a set of computer building blocks called Macromodules was undertaken during the late 60's and early 70's. An inventory of approximately 1000 modules was completed. The primary goal of these modules was to simplify the task of building experimental computer systems. The design of a computer system consisted of drawing a flow chart, then plugging modules into a structure which supplied power, cooling, and most connections, and finally plugging in cables in one-to-one correspondence with the flow chart. Molnar was the primary innovator in the electrical design which eliminated the possibility of electrical or timing errors by users of the modules; any connection that was mechanically possible was electrically correct. Many experimental computer systems were built from these modules, in particular for molecular graphics, drug design, and detection on cardiac arrhythmias from electrocardiograms.

The Macromodules used asynchronous, or self-timed, logic in order to eliminate the possibility of timing errors. Much of Molnar's later work on asynchronous circuits was inspired by the Macromodule effort. His goals were to develop systematic and rigorous design methods that could be used to produce provably correct asynchronous circuits and systems. His extended visits to Caltech and Eindhoven Institute of Technology were to advance the theory of asynchronous circuit design through collaborations with Charles L. Seitz, Martin Rem, and many others. He was cofounder with Martin Rem of a series of workshops on asynchronous circuit design, the most recent of which was Async97.

Molnar was one of the very first to recognize the potential for error ("Science and the Citizen" Scientific American, Vol. 22, pp. 43-44, Apr. 1973.) in conventional clocked computer systems from metastability, and worked to make this problem more widely recognized and better understood. He had been intrigued by rare "glitches" in the operation of the LINC, which he tracked down to subtle interactions between the timing of changing external inputs and the program clock. He and colleagues subsequently demonstrated that the glitch or metastability problem is fundamental in nature and can only be reduced probabilistically, not entirely eliminated. As several of the causes he championed, this was derided by some as "not a real problem," but as usual he persevered and his ideas were eventually widely accepted.

Molnar established the Institute for Biomedical Computing in 1983 to foster the joint interests in biomedical computing of the Schools of Engineering and Medicine. He devoted major effort to bridging the cultural gap between the schools and was ideally qualified to do so based on his fundamental research ranging from computer design to auditory physiology. Beginning with the LINC, and throughout his career, Molnar catalyzed the spread of computer tools to biomedical researchers.

In 1990 he began consulting with Sun Microsystems in California, and in 1995 he left Washington University to pursue his interests in asynchronous systems and design methodology as director of the Science Office at Sun Microsystems Laboratories, where he collaborated with Ivan Sutherland and Bob Sproull in a major research effort on asynchronous computers. Molnar's pioneering work on asynchronous circuit design has become increasingly relevant as designers face problems such as modular and hierarchical design, and extremely high power dissipation, and as more and more research groups and laboratories attempt to exploit the advantages of asynchronous circuits.

Charlie was not a prolific publisher. His striving for perfection made it difficult to for him to publish because by the time any work was reduced to words and equations on paper, he had new ideas that obsoleted what had been written. His "Red Pen" was famous; he made major contributions to the style, clarity, content, and organization of many papers and other publications for which he did not ask, and frequently did not receive, acknowledgement. However, any paper or publication that he touched was improved dramatically. He would sometimes comment: "mean what you say, say what you mean." He much preferred one-on-one discussions to writing or to formal lectures; his discussions were frequently accompanied by long walks. Somehow, talking was improved when carried out in conjunction with walking. His primary legacy will be his ideas, and the principles of integrity, style, conciseness, and provable correctness that he instilled in his colleagues, students, and associates.

Quotations:

"What's the point of having principles if you don't use them when the going is tough? That's what you're paying them for!"

"Ankle deep, head down."

"How do you know it's right?"

Significant Publications:

Clark, W.A. and C.E. Molnar, 1964, "The LINC," Anal. New York Academy of Sciences, Vol 115, pp. 653-658.

Clark, W.A. and C.E. Molnar, 1965, "A Description of the LINC," Computers in Biomedical Research, Vol II, B.D. Waxman and R. Stacey, eds, Academic Press, New York, NY.

Model for the Convergence of Inputs Upon Neurons in the Cochlear Nucleus, D.Sc. Thesis, MIT, 1966.

Chaney, T.J. and C.E. Molnar, "Anomalous Behavior of Synchronizer and Arbiter Circuits," IEEE Trans. on Computers, Vol. C-22, No. 4, pp. 421-422, Apr. 1973.

Brief excerpts of video and audio from a series of two lectures on metastability given at Hewlett Packard, February, 1992: In Preparation.

Clark, W.A. and C.E. Molnar, 1974, "Macromodular Computer Systems," Computers in Biomedical Research, pp. 45-85, Vol IV, B.D. Waxman and R. Stacey, eds, Academic Press, New York, NY.

Kim, D.O. and C.E. Molnar: Cochlear mechanics: Measurements and models, in The Nervous System, Vol. 3, Human Communication and Its Disorders, edited by DB Tower (Raven, New York) 1975; pp 57-68

Sproull, R.F., I.E. Sutherland, and C.E. Molnar, 1994, "The Counterflow Pipeline Architecture," IEEE Design and Test of Computers, Vol. 11, no.3, pp. 44-59.


Back to Molnar home page.

This page last updated: 2002.11.05 by fred@cse.wustl.edu